MissingnessAudit.py

# -*- coding: utf-8 -*-
"""MissingnessAudit.ipynb

Automatically generated by Colab.

Original file is located at
    https://colab.research.google.com/drive/10ktqYR6Cv7gMByA9WSlIqMg-NWywhRU_
"""

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns
file_path = "DigitalNomadPolicyDataSet.xlsx"
xls = pd.ExcelFile(file_path)
print(xls.sheet_names)

data = {
    sheet: pd.read_excel(file_path, sheet_name=sheet)
    for sheet in xls.sheet_names
}

# Variable-level missingness
for sheet_name, df in data.items():
    missing_summary = pd.DataFrame({
        "Variable": df.columns,
        "Missing_Count": df.isna().sum().values,
        "Missing_Percent": np.round(df.isna().mean().values * 100, 2)
    })
    missing_summary = missing_summary.sort_values(
        "Missing_Percent",
        ascending=False
    )
    print("\nTop variables with missing values:")
    print(missing_summary.head(20))
    missing_summary.to_csv(
        f"{sheet_name}_missingness_summary.csv",
        index=False
    )

#Dataset-level missingness
for sheet_name, df in data.items():
    total_cells = np.prod(df.shape)
    missing_cells = df.isna().sum().sum()
    print(f"\n{sheet_name}")
    print(f"Rows: {df.shape[0]:,}")
    print(f"Columns: {df.shape[1]:,}")
    print(f"Total Missing Cells: {missing_cells:,}")
    print(f"Overall Missingness: {100*missing_cells/total_cells:.2f}%")

# Country-level missingness
for sheet_name, df in data.items():
    if 'iso3' in df.columns:
        country_missing = (
            df.groupby('iso3')
            .apply(lambda x: x.isna().mean().mean()*100)
            .reset_index(name='Missing_Percent')
            .sort_values('Missing_Percent', ascending=False)
        )
        country_missing.to_csv(
            f"{sheet_name}_country_missingness.csv",
            index=False
        )

        print(f"\nTop countries with missing data ({sheet_name})")
        print(country_missing.head(10))

# 5. Year-level missingness

for sheet_name, df in data.items():

    if 'year' in df.columns:

        yearly_missing = (
            df.groupby('year')
            .apply(lambda x: x.isna().mean().mean()*100)
            .reset_index(name='Missing_Percent')
        )

        yearly_missing.to_csv(
            f"{sheet_name}_year_missingness.csv",
            index=False
        )

        plt.figure(figsize=(10,5))
        sns.lineplot(
            data=yearly_missing,
            x='year',
            y='Missing_Percent'
        )
        plt.title(f"Missingness by Year: {sheet_name}")
        plt.ylabel("% Missing")
        plt.tight_layout()
        plt.savefig(
            f"{sheet_name}_yearly_missingness.png",
            dpi=300
        )
        plt.close()

# --------------------------------------------
# 6. Missingness Heatmap

for sheet_name, df in data.items():

    plt.figure(figsize=(14,8))

    sns.heatmap(
        df.isna(),
        cbar=True,
        yticklabels=False
    )

    plt.title(f"Missing Data Pattern: {sheet_name}")

    plt.tight_layout()

    plt.savefig(
        f"{sheet_name}_missingness_heatmap.png",
        dpi=300
    )

    plt.close()

# Data Descriptor Table

for sheet_name, df in data.items():

    audit_table = pd.DataFrame({
        "Variable": df.columns,
        "Data_Type": df.dtypes.astype(str),
        "Missing_Count": df.isna().sum(),
        "Missing_Percent": round(df.isna().mean()*100,2),
        "Unique_Values": df.nunique()
    })

    audit_table.to_excel(
        f"{sheet_name}_audit_table.xlsx",
        index=False
    )